Skip to main content
Mako
Mako is a 32-billion parameter language model fine-tuned from Qwen3-32B. She has her own personality, 16 built-in tools, and read-only access to 8 EVM blockchains. She’s not an assistant. She’s an agent.

Quickstart

Make your first API call in 30 seconds.

Architecture

How the gateway, model backends, and tool system fit together.

On-Chain Tools

Balances, tokens, ENS, gas, transactions — across 8 chains.

Self-Host

Run the full Mako stack on your own infrastructure.

​
Agentic tool orchestration

Mako chains tools together autonomously. Ask a compound question and she figures out the steps: 
User: "what's vitalik.eth's balance worth in usd?"
Under the hood, Mako executes:
  1. resolve_ens("vitalik.eth") → 0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045
  2. get_eth_balance("0xd8dA...") → 5.69 ETH
  3. get_crypto_price("ethereum") → $1,665
  4. Math → “vitalik.eth has about $9,480 in eth”
All in a single API call. No tool-call handling on your end — the gateway orchestrates everything server-side and returns the final answer.

​
Supported chains

Every on-chain tool accepts an optional chain parameter. Default is ethereum.
ChainAliasesRPC
Ethereumethereum, ethethereum-rpc.publicnode.com
Basebasemainnet.base.org
Arbitrumarbitrum, arbarb1.arbitrum.io/rpc
Optimismoptimism, opmainnet.optimism.io
Polygonpolygon, maticpolygon-rpc.com
Avalancheavalanche, avaxapi.avax.network/ext/bc/C/rpc
BSCbsc, bnbbsc-dataseed.binance.org
Gnosisgnosisrpc.gnosischain.com

​
The API is OpenAI-compatible

Drop Mako into anything that speaks the OpenAI Chat Completions format: 
from openai import OpenAI

client = OpenAI(
    base_url="https://gateway.deepmako.com/v1",
    api_key="your_api_key",
    default_headers={"x-wallet-address": "0xYourAddress"}
)

r = client.chat.completions.create(
    model="conductor",
    messages=[{"role": "user", "content": "gas price on base?"}]
)
print(r.choices[0].message.content)
# "base gas is at 0.0118 gwei right now. basically free."

​
Open source

The full stack is open source at github.com/DeepMako/mako:
  • Gateway — Express.js API server with tool orchestration, auth, and credits
  • Frontend — Next.js chat interface at deepmako.com
  • Model weights — Fine-tuned Qwen3-32B (Conductor) and 8B (Operator)
  • Knowledge base — Curated JSON entries for 15+ Base ecosystem projects